Red Hat Bugzilla – Bug 1222025
unset arguments are not treated as an error when set -u is used
Last modified: 2017-08-01 12:26:55 EDT
Description of problem: When set -u is used, ksh should treat unset parameters as an error when substituting. This does not work for unset positional parameters like $1. It does work, however, for usual variables (see the attached reproducer). Version-Release number of selected component (if applicable): 20120801-22.el7_1.1 How reproducible: always Steps to Reproduce: 1. run the attached reproducer Actual results: No error printed for $1. Expected results: An error printed for $1 Additional info: N/A
Created attachment 1025897 [details] reproducer
This is intentional change made by upstream. 09-05-12 To conform with POSIX, the -u option only checks for unset variables and subscript elements rather than checking for all parameters.
Michal, Had the customer ask the following: "Your update is clear that upstream made this as a deliberate change to conform with POSIX. Is the upstream position that this is not a bug/ regression and will not be fixed and that the previous behaviour will not be made available through an option/ environment variable? Tim " I re-opened this so that you'll get the notification. I would like to have an answer to this as I think it's a really well thought out question. Cheers, Karl
Michal, Also: "Hi Karl, A point of interest: Checking with our Solaris team they have a case open with Oracle for the same issue. Oracle responded to our Solaris team today that they (Oracle Solaris engineering) have been working on a change which has been integrated into the upstream. Tim" Karl
When the change in varsub() in macro.c mentioned in the following post is made, we get in the newly added part of the code, when $1 is being expanded: http://lists.research.att.com/pipermail/ast-developers/2015q2/004038.html Then strchr() seems to return a pointer to the terminator and errormsg() is not called.
Upstream patch is slightly different from the submitted patch in comment 12, check for non-zero d is unnecessary : https://github.com/att/ast/blob/beta/src/cmd/ksh93/sh/macro.c#L1230-L1236 if(!v && sh_isoption(mp->shp,SH_NOUNSET)) { d=fcget(); fcseek(-1); if(!strchr(":+-?=",d)) errormsg(SH_DICT,ERROR_exit(1),e_notset,ltos(c)); }
strchr() can search for NULL characters in string, so check for 0 is required. The patch in comment 12 looks good.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2017:1936